Qwen: Qwen3 32B
qwen/qwen3-32b
Cheapest provider
$0.08 / 1M
DeepInfra
Fastest provider (p95)
—
No throughput data yet — populated as traffic accumulates
Intelligence (composite)
70.9
MMLU-Pro · HumanEval · math · GPQA
Per-provider performance
Latency / throughput / uptime / price measured across providers over the last 30 minutes of live traffic. This is what proves “sourced cheapest” — Atlas mode draws on these per call to serve the cheapest path that holds quality.
| Provider | Quant | Input $/1M | Output $/1M | Latency p50 / p95 | Throughput p50 / p95 | Uptime 30m | Success |
|---|---|---|---|---|---|---|---|
| DeepInfra | q8· fp8 | $0.0800 | $0.2800 | — | — | 99.99% | — |
| AtlasCloud | q8· fp8 | $0.1000 | $1.2000 | — | — | 99.99% | — |
| Nebius | q8· fp8 | $0.1000 | $0.3000 | — | — | 100.00% | — |
| Novita | q8· fp8 | $0.1000 | $0.4500 | — | — | 85.88% | — |
| Alibaba | undisclosed | $0.1040 | $0.4160 | — | — | 99.99% | — |
| SiliconFlow | q8· fp8 | $0.1400 | $0.5700 | — | — | 90.24% | — |
| Groq | undisclosed | $0.2900 | $0.5900 | — | — | 99.96% | — |
“—” means live telemetry hasn’t accumulated enough recent traffic for that endpoint. “undisclosed” means the provider serves the model but doesn’t expose the quantization label (typically running fp8 / int8 internally).
Intelligence breakdown
Composite score is a weighted average of public benchmarks (30% MMLU-Pro, 25% code pass@1, 25% math, 20% GPQA). Numbers come from model cards and the Artificial Analysis intelligence harness; missing components are renormalised over what’s present.
MMLU-Pro
—
broad reasoning
Code
—
pass@1 (HumanEval / LiveCodeBench)
AIME 2025
72.9
math accuracy
GPQA Diamond
68.4
hard reasoning
Source: Qwen3-32B (GPQA-Diamond 68.4, AIME25 72.9); DeepSeek-R1-0528 model card comparison table
How Atlas mode sources Qwen: Qwen3 32B
- Strict mode — pin Qwen: Qwen3 32B exactly and we pass it straight through, sourced from the cheapest provider above. The same model, no substitutions — currently DeepInfra at $0.08/1M.
- Atlas mode — the default. Each call is auto-optimized for the cheapest path that holds quality, at least 5% off going direct from call one and climbing as it ramps. You always see which model served the call and exactly what you saved — thumbs-down anything you don’t like for a full refund.